Article image
Here map

The cars of the future will do their own planning

A self-driving car can steer itself using its sensors, but also needs an overall strategic plan. This is something that HERE Technologies is working on with its latest mapping technology where the map helps the sensors put the car in context in real time. 
"Eventually, these systems will be so good that they'll be able to assess every situation on the roads," says Petter Djerf at HERE in Gothenburg.

Main content

"If you consider how maps are developing, digital maps have taken over completely, and have almost become a standard in new cars and a given on mobile phones, which have multiple map apps. They're almost a natural part of websites where you can track logistical flows, find an office and directions and so on." 

Petter Djerf, senior account executive at HERE Technologies, has worked with digital maps and telematics since the late 1980s when he was a commissioned officer in the Swedish navy. After ten years of telecom systems and navigation in a marine environment, he came ashore and began a civilian career that revolved around traffic information, logistics and transport before joining HERE Technologies eighteen months ago.  

HERE Technologies provides data mapping and related services. The company was previously owned by Nokia and at that time was focused on consumer-oriented maps for Nokia mobile phones, but is now owned by the three German auto manufacturers Audi, BMW and Mercedes, together with Intel and a number of Chinese investors.    

Image
Petter Djerf

"About 9 out of 10 cars use HERE's maps. One trend is that maps are everywhere, another is that now they are being integrated with increasingly smaller devices. In the same way that we take for granted that we can locate our car in a car park or use Volvo's In-car Delivery service to deliver goods to our car wherever it's parked, location services are becoming an integral part of other, increasingly smaller devices. We're also seeing a very interesting trend in the automotive industry, which is for the car to increasingly take over the map reading role. It's almost the case in some modern cars that the map for the driver is not quite as important as the map for the car."

Today, auto manufacturers use so-called ADAS (advanced driver assistance systems) maps to control the advanced driver assistance systems. They provide a reliable model of the terrain with altitudes and bends and with these as a basis you can set the car's systems to provide quite a comfortable ride. But there are very many other things in the car's surroundings that also need to be incorporated in the map.

"There's a big difference between a 90-degree bend on the plains of southern Sweden and the same bend along a mountain road in Croatia – you'd rather skid into a field in Sweden than a ravine in Croatia. However, that it feels much safer to drive a little faster on the plains of Sweden is not something that gets included in this standard bend mapping. Instead we have to add something else: what would a human do? We need to learn to integrate this data and provide our vehicles with the right attributes so that the very smart digital map with a myriad of profiles can together with the car's sensors create a pleasant driving experience."

A self-driving car can control itself well and keep to the road with the help of its maps and sensors. It knows there's a 70 km/h speed limit, whether the centre line is broken or solid, how wide the road is and so on. But it's also important that the car has an overall strategic plan for the journey. This is something that HERE is working on with its latest mapping technology, machine-to-machine maps, where the map helps the sensors put the car in context in real time.  

"Say that you're driving north on the E6 and want to exit at Kallebäck. If you have a self-driving car and program it to always seek out openings in the traffic, maybe it will make its way over to the left lane. But if there's heavy traffic heading for the Kallebäck exit your car should instead start to move into the right lane before it's too late without disrupting other traffic. This is something that driverless cars must be able to manage themselves and that's when real-time information about traffic jams, rain, ice, roadworks, accidents and so on is an extremely important data layer. If the information indicates a difficult situation, the car can return control to the driver. Eventually, these systems will be so good that they'll assess everything, based on the maps and with the aid of their own sensors." 

HERE Technologies has had hundreds of cars on the roads essentially all over the world and has mapped the physical reality with laser radar, such as the basic structures of uphill gradients, downhill gradients and bends.  

"What we're doing now is to document the roads so that we have accurate lane descriptions with hard shoulders, solid centre lines, widths and so on that we can include in the maps. What's more, we add objects that can be used to get a fix on the car's position, you could call it a reference system with an unbelievable number of triangulation points – road signs, utility poles, electrical enclosures, overpasses. By continually verifying these objects, the car can adjust its position on the fly. It's very similar to traditional marine navigation…"

Another trend that HERE Technologies has a great deal of faith in is big data. In HERE's case, this means gathering the sensor data from the vehicles.

"We've already linked our owners' infrastructures and this amount of information can be used for so much – different types of weather values, road surface friction values, parking services. With the aid of the sensor data, we're starting to amass a great deal of information that will help us build a better map. Things are moving very quickly now, just how fast depends on which companies join us and which sensor data they choose to share with us. However, there are a lot of legal and privacy issues to deal with, so it will take time, even though technically speaking it's not difficult."  

What is required to manage the legal aspects?

"I've always thought that the simplest solution is for major companies to try to be open to the future. Auto manufacturers ought to have told their users ten years ago that you and everyone else who buys this car will be able to be tracked by our technical systems – and you approve this, otherwise you can't make full use of the car. Something like that. Then we would've been done with it. Then there are many other legal issues that are extremely complex, such as who is responsible if a driverless car hits one of my children. We also need a solution to this and I don't feel that we have one. When it comes to the technology, the difficulty is in standardisation. For instance, it's extremely important that we develop a global standard for sensor data." 

How does the smart city fit into your business?

"Even though historically we've worked very closely with the automotive industry, we have a unique position thanks to our high-resolution maps and all the thousands of geotechnical devices continually providing new map data. Naturally, the maps can be used in very many other areas – where smaller objects are located, transport and logistics management, trade and much more. One current trend is to link your car's infotainment system to your Outlook mail and calendar. You're sitting in your car somewhere and you're starting to run low on fuel. You usually refuel at the station there and the car suggests that you do so now and hold your next meeting over the phone with the following Skype details and you really just select OK, OK, OK with the steering wheel. You're guided to the filling station, the call is connected and you have your meeting. Or your car sees that your meeting at a particular location starts in 10 minutes but you're half an hour away, so the car asks if you want to send an email to the meeting organiser to tell them that you're running late. Integrating all this stuff with the car to make it smart is going to be big business. Then we have the car sharing trend and the more you tie all these things together, the smarter our cities will become. It's not science fiction as these things are already pretty smart today – and we can make them smarter." 

What do you think of the Gothenburg region's automotive cluster?

"In terms of knowledge, we're incredibly advanced, not least in terms of expertise per capita. At the same time, I think – without sticking my neck out too much – that sometimes we're overconfident. When you see fully automated transportation in Singapore or self-driving cars in Lyon or Paris, which are actually up and running right now, then it doesn't feel like we're light years ahead of everyone else – we should adopt a more humble approach and incorporate more benchmarking. If we as an industry fail to challenge ourselves enough, there's a pretty big risk that we'll see an accelerating rate in the establishment of new Ubers and Teslas – the real game changers – elsewhere, and then entire structures could collapse. This is why it's extremely important that we have organisations like Telematics Valley in Lindholmen where we regularly invite other people and dare to share information across corporate borders and reveal what we're doing. This is something that I feel we're very good at in Gothenburg."

Print this page:

Tip a friend

Bottom content